Cloud infrastructure and developer tools for web applications and AI agents
Vercel
Vercel provides cloud infrastructure and developer tools for building and deploying web applications and AI agents. Developers can host frontends, backend APIs and container workloads, while v0 turns prompts into software. Its Agent Stack connects model access, execution and integrations: AI SDK standardizes streaming, tool calls and structured output across model providers; AI Gateway routes requests through one endpoint with provider failover. Workflow SDK adds durable execution with retries and persisted state, while Vercel Sandbox gives agents isolated microVMs to run and test generated code.
Founded in 2015 by Guillermo Rauch, Tony Kovanen and Naoyuki Kanezawa, the company is led by CEO Rauch. Originally called Zeit, it became Vercel in 2020. Its enduring engineering contribution is Next.js, introduced as an open-source React framework in 2016. Next.js brought filesystem routing, automatic server rendering and code splitting together, helping developers build applications that share JavaScript between client and server.
In June 2026, Vercel reported more than 16 million weekly AI SDK downloads. It also reported over 500 million Next.js downloads in the 12 months preceding its September 2025 financing announcement. That $300 million Series F, co-led by Accel and GIC, valued Vercel at $9.3 billion.
6 talks
Newest first
5 speakers at AIE
Affiliations reflect their AIE appearances, not necessarily current employment.
Start here
- Vercel AI SDK Masterclass: From Fundamentals to Deep Research
Start with Nico Albanese’s workshop to learn how text generation, tools, and structured outputs combine into a recursive research workflow that produces a report.
Nico AlbaneseAI Engineer Summit 2025
- Building durable Agents with Workflow DevKit & AI SDK
Use Peter Wielander’s workshop to understand how workflow and step directives support webhooks, resumable streams, and human participation in agent execution.
Peter WielanderAI Engineer Code 2025
- Evals Are Not Unit Tests
Watch Ido Pesok’s demonstration to understand why a few successful manual tests cannot establish reliability for a nondeterministic application.
Ido PesokAI Engineer World's Fair 2025
- The Weekend AI Engineer
Hassan El Mghari explains how he built open-source AI applications on weekends and attracted millions of users.
Hassan El MghariAI Engineer Summit 2023
Messages from the stage
Give agents state and resumable execution
Nico Albanese demonstrates accumulated research state and isolated working files. Peter Wielander adds checkpoints, suspended execution, and operational visibility to a coding-agent workflow.
Test against real input variation
Ido Pesok uses a fruit-letter-counting application to show how unexpected production inputs expose failures. He recommends representative user examples, reusable evaluation data, and evaluations in CI.
New economics for small applications
Hassan El Mghari describes rapid experimentation with readily available AI APIs and organic distribution. Malte Ubl argues that agents make previously uneconomical software practical and increase the importance of APIs and CLIs.
Affiliations reflect each recorded session, not necessarily current employment.